Deep hybrid modeling of a cell culture bioreactor data using Long Short-Term Memory (LSTM) networks combined with first principles equations
To run hybrid model training or simulation of two previously trained hybrid model use the code "hybnet_train_main.m" This codes shows how to define model structures and which parameter to use to run the training process. Furthemore, several ways to plot and analyse the results
This code "hybnet_train_main.m" comes with two predefined hybrid model structures one Feed Forward Neural Network (FFNN) and one Long Short-Term Memory (LSTM). Both structures were pre-trained and the data saved in "hybrid_FFNN_1.mat" and "hybrid_LSTM_1.mat". It asks if the user wants to train these model structures again or simulate from saved files. Different plots are generated and a excel file "structures_fit_results.xlsx" with the overall results.
The folder ~/data contains data.xlsx with the feed DoE details and the simulations of concentrations over time generated using a dynamic model. This is a synthetic dataset, the model was created based on the metabolic model proposed by Robitaille et al. (2015). It also contains "data.mat" which is the import the concentrations, feed and other relevant informations. This file is generated using "/data/main_data_processing.m". When imported into matlab data(i).accum is the total amount of a metabolite that should be in the bioreactor over time, it is the "sum of all added concentrations × volume added - sample volume × reactor concentration". The file also contains data(i).m_r which are the reacted amounts over time. The calculation of data(i).accum is made during the process of data(i).m_r calculation. The latter is described in the supplementary material of this paper or the file "/data/read_me_data_processing.doc"
Data for the paper: Deep hybrid modeling of a HEK293 process: combining Long Short-Term Memory (LSTM) networks with first principles equations João R. C. Ramos, José Pinto, Gil Poiares-Oliveira, Ludovic Peeters, Patrick Dumas, Rui Oliveira